You are on page 1of 14

Petroleum and Coal

Article Open Access

EXPERIMENTAL STUDY OF HIGH LIQUID VISCOSITY OIL-GAS FLOWS USING ECT AND GAMMA
RADIATION METHODS

Archibong Archibong-Eso1,2, Yahaya D. Baba1,3*, Aliyu Aliyu1,4, Nonso Okeke1, Abdulwahab


Giwa3
1 Oil& Gas Engineering Centre, Cranfield University, UK
2 Cross River University of Technology, Calabar, Nigeria
3 Department of Chemical and Petroleum Engineering, Afe Babalola University,Ado-Ekiti, Nigeria
4 School of Mechanical Engineering, Pusan National University, 609-735, Busan, Republic of Korea

Received October 19, 2017; Accepted January 8, 2018

Abstract
Electrical capacitance tomography (ECT) and gamma ray densitometry have found applications in
multiphase flows as both have been used to visualize flow patterns in the process, nuclear, chemical
and oil-gas industries. However, liquid viscosity is often limited to about 10 cP in most of these
applications. This study evaluates the suitability of ECT in the monitoring of multiphase flow consisting
of heavy-oil (up to 7500 cP) and gas. Heavy-oil is of interest because of its huge reserves, increasing
world energy demand, dwindling conventional-oil reserves, etc. The results obtained showed that ECT
performed well in flow visualization and measurement of high-viscous oil/gas flows. Flow patterns
identified included plug, slug, pseudo-slug (blow-through-slug) and annular flows. Liquid holdup
measurements were found to be within ±10% and ±5% error in the static stratified and static annular
test, respectively, based on the LBP algorithm while in dynamic test, it was considerably dependent on
flow patterns and the LBP. Comparatively, high definition video recordings, pictures and gamma rays
densitometer were used to validate the flow patterns identified by ECT.
Keywords: High viscosity oil; flow patterns; gamma rays; multiphase flow; liquid holdup.

1. Introduction
The foray of tomography into chemical and process industries is hinged on its several
successes in medical applications. Multiphase flows exist widely in process and chemical
industries such as nuclear, food, agro-allied, oil and gas, water, polymer, textile, etc. Prior
knowledge of the behaviour of the phase distribution plays a significant role in process control,
process equipment design, pumps and pumping systems design, quality control, safety assu-
rance, process flow assurance, etc. The complex nature of multiphase flow, which is charac-
terised by deformable phase interface, phase inversion, phase slip, as well as the variations
of the phase distribution and the velocity profiles with temporal and spatial coordinates, makes
it very difficult to measure its phase quantity and distribution (flow pattern or flow regime).
Tomography thus becomes important as it can be utilized as a valuable tool in flow visua-
lization. Of the various industrial tomography such as gamma, X-ray, microwave, ultrasonic,
optical, electrical resistance tomography (ERT), electrical capacitance tomography (ECT) that
are available for process industry applications, ECT is one of the most promising one due to
its inherent simplicity, robustness, high speed capability, zero radiation emission, non-inva-
siveness, non-intrusiveness, low cost and ability to withstand wide range of operational tem-
peratures and pressures [1].
ECT is a tomographic technique that involves the determination of the spatial permittivity
distribution of dielectric materials within an object from external capacitance measurement
using multiple sensor electrodes mounted on the periphery of the object [2]. The ECT system

Pet Coal (2018); 60(2): 196-209


ISSN 1337-7027 an open access journal
196
Petroleum and Coal

typically consists of three units, namely the ECT sensing system, Data Acquisition System
(DAS) and the Computing System. The ECT sensing system consists of an object (e.g., a pipe)
and sensing electrodes, which are mounted on the object’s periphery; this unit is used to
obtain measurement values of the dielectric material within the object where they are mounted.
Data Acquisition System (DAS) unit is used to acquire data obtained by the sensors and
subsequently transfer the data in machine-readable form to the computing system. The
computing system consists of a computer built to ECT manufacturer’s specification and pro-
prietary software, which comprises algorithms that aids real time and offline data recons-
truction, tomographic image display and data analysis.
The ECT system was reportedly used in pipeline flow visualization at the University of Man-
chester in the early 1990s [3]. The success of earlier studies led to the subsequent use of ECT
in visualizing multiphase flows in oil pipelines [1]. Nevertheless, visualization of multiphase
flow using ECT techniques were concentrated on low viscous oils (less than 20 cP). Research
work using ECT for applications higher than 1000 cP was not found in literature by the authors.
Also, in multiphase flows, flow patterns for flows consisting of high-viscous liquids are signi-
ficantly different from low-viscous liquids. Gokcal et al. [4] observed an increase in slug frequency
and liquid holdup with a corresponding decrease in slug length as oil viscosity was increasing.
Márquez and Trujillo [5] affirmed the difference in flow pattern at increased liquid viscosity
when they noted in their work that slug region increased with increase in liquid viscosity. This
study was concerned with high-viscous oil applications due to its increasing importance as a
veritable energy source. Factors such as depleting conventional oil reserves, increase in world
energy demand, technological advances (in heavy oil production), availability of huge heavy
oil reserves and prevailing oil prices have combined to make heavy oil one of the leading sources
of energy. Previous studies of ECT on multiphase applications are outlined below.
Gamio et al.[6] used an ECT system they designed to visualize two-phase flow consisting of
oil and gas in a 7.62 cm pressurised test loop. Nitrogen gas and Exxsol D80 oil were respect-
tively used as the gas and oil phases in the study. Different flow patterns were obtained though
variations in the flow rates of oil and gas. Tomographic images were obtained for each flow
condition and the results obtained were validated with video graphic images obtained through
a transparent section in the flow loop as well as with the flow pattern map. Their work was based
on oil with viscosity of 1.7 cP at 25 oC, and the study did not involve flow measurement such
as liquid holdup.
Zhang and Wang [7] used a twin-plane dual modality (ECT/ERT) system developed by
Tianjin University for the identification of oil-gas two-phase flow patterns. Only the ECT sensor
outputs were used and the ECT sensor consisted of 12 electrodes mounted at the circumfe-
rence of the pipe. A 0.6 m long horizontal test section with pipe internal diameter of 80 mm
was used for the study. Nylon plastic pellets were used to stimulate each flow pattern. In the
work, ECT identification of stratified flow, annular flow, core flow and full pipe flow patterns
agreed with the Support Vector Machine (SVM), which is a technique based on statistical
theory. However, this study reported very limited experimental data. The test was done under
static conditions and the authors did not list the properties of the fluids used.
Baker et al. [8] used ECT to visualize the time varying cross sectional distribution transients
in two-phase gas-liquid stratified/slug flow. The experiments were conducted in a 16.7 m long
pipe with internal diameter of 0.038 m. Air and kerosene (viscosity of 3 cP at 25 oC) were
used as the gas and liquid phase respectively. The maximum gas superficial velocity used was
12 m/s while the superficial velocity of the liquid was 0.5 m/s. An 8 electrode ECT sensor was
used to visualize the cross sectional distribution of the phases in flow. Stratified and slug flows
were clearly shown in the tomographic images obtained for both the high and low liquid
superficial velocities. However, ECT images and visual observations via the transparent pipe
section indicated either slug or stratified flows in disparity to annular flow predicted by the
flow pattern map of [9]. Void fractions data obtained from ECT were found to be in agreement
with those obtained from other studies in literature with similar fluid properties.

Pet Coal (2018); 60(2): 196-209


ISSN 1337-7027 an open access journal
197
Petroleum and Coal

Isaksent et al. [10] used capacitance based tomography system for interface measurement
in separation vessels. An eight-electrode capacitance sensor mounted circumferentially around
a plexi-glass separator with a 60 cm long earthed screen surrounding the electrodes was used.
The test fluids used were compressed air, tap and salt water as well as two types of oil (Exxol
D80 and D100) with dielectric constant near 2.1. The tomography system was tested for both
oil/air separation and oil/water/air separation. The oil/foam/air was conducted by creating
enough foam inside the separator by means of high oil and gas flow rates which resulted in severe
liquid motion. The tomographic system detected the foam thickness inside the separator above
5 to 10 cm. The reconstructed foam layer thickness was observed to give a good represent-
tation of the foam layer thickness inside the separator. The average absolute and maximum
absolute errors were 0.3 and 2 cm, respectively. Water-oil-air test was also carried out for
water conductivities of 4x10-3 S/m and 5 S/m (salt water). The absolute average error for
both tap and salt water height and oil thickness were within 0.5 and 1.0 cm respectively.
Warsito and Fan [11] used a capacitance sensor array that comprised a twin plane sensor
using 12 electrodes for each plane with each electrode measuring 5 cm in length to identify
flow patterns in a gas-liquid two phase flow and a three phase gas-liquid-solid flow in a bubble
column. For the gas-liquid fluid, air was used as the gas phase while Norpar 15 (paraffin) and
Paratherm heat transfer liquids were used as the liquid phase. Norpar 15 has a density of 773
kg/m3 and a viscosity of 0.253 mPa∙s. Paratherm has a density and a viscosity of 870 kg/m3
and 0.317 mPa∙s respectively. Polystyrene beads with density and diameter of 1020 kg/cm 3
and 2 mm, respectively, were used as the solid phase in the gas–liquid–solid system. The three-
phase system was essentially a dielectrically two-phase system, enabling the measurement of
the gas holdup alone in the gas–liquid–solid system independent of the other two phases. This
is because the polystyrene beads have permittivity similar to that of Paratherm. The data
acquisition system captured data at 100 frames per second. Iterative Linear Back Projection
(ILBP) algorithm was used for the image reconstruction of the multiphase flow at 32 x 32
pixels per image. The study showed that ECT could be suitably applied in imaging gas-liquid
as well as gas-liquid-solid flow systems with the image reconstruction method developed by
the authors. It was observed that the technique used had no limitations on the maximum gas
holdup but it proved difficult to measure small gas holdup due to the small signal-to-noise ratio.
Jeanmeure et al. [12] developed a design that enabled the use of ECT to directly identify
flow patterns in a two-phase gas-oil flow. The study involved the flow identification and control
in an air-kerosene flow in horizontal piping by using 8-electrodes ECT system and T-junction
separator. This was done for a reduced set of interest of flow patterns with distinctive charac-
teristics (annular and stratified flow). The method produced relatively fast images at 200
images per second and was used to detect slug flow to enable actions that will prevent
undesirable effects on unit operation equipment downstream. The density and viscosity of the
oil used in the test were 973.05 kg/m3 and 0.026 Pa∙s respectively. The flow pattern identi-
fication was performed using the visual inspection of the flow in transparent pipe section and
quantifying the fluctuations of the void fraction and dynamic pressure from the capacitance
reading and pressure readings respectively using Probability Density Function (PDF) and Power
Spectral Density (PSD) function of the time trace signal. The study concluded that capacitance
sensors were very sensitive to flow regime change and could be used to identify the different
flow patterns. This study was, actually, on low viscous oil, and tomographic images were not
reported there.
Similarly, gamma radiation methods, being non-intrusive methods, have been frequently
used for the measurement of component volume fractions in multiphase flow systems. For
most applications, it is the attenuation of radiation (gamma-ray) that serves as the source for
the measurements. Gamma densitometry has numerous advantages in terms of its higher
penetration capabilities over other radiation attenuation methods (i.e., neutron beams) thereby
making it a more ideal system for the measurement of phase fractions in large industrial
systems.

Pet Coal (2018); 60(2): 196-209


ISSN 1337-7027 an open access journal
198
Petroleum and Coal

From the literature reviewed above, it is obvious that most existing studies were based on
low viscous oil, hence, a basis for this study to consider high-viscous flows using ECT. The
capability of ECT in high-viscous flows (up to µ=7500 cP) was thus investigated under static
(bench) and dynamic (flowing) conditions. Tomographic images of ECT and temporal varia-
tions of liquid holdup measurements through time series plots were used to identify flow
patterns. Gamma densitometer (a device for measuring the density of fluids in a completed
well, pipeline or process equipment, using a radioactive source of gamma rays and a detector),
visual observations and video recordings were used to validate the measurements of the ECT.
2. Experimental setup
In this study, two experimental test facilities in the Process and Systems Engineering
(PASE) Laboratory were used for the dynamic (flowing) test, and the bench test was also
conducted. Actually, 1 and 3-inch multiphase facility are similar in design with both having
capabilities to operate under varieties of multiphase flows consisting of gas, water, oil and
sand. The 3-inch multiphase facility (Figure 1) that was used in this work is described below.

Figure 1. Schematics of the 3-inch multiphase facility at Cranfield University


The 3-inch multiphase facility is similar to the 1-inch facility. It is, actually, a scale up of
the 1-inch facility. It is once through and consists of a 0.75 mm ID Perspex pipe with length
of about 17 m. It has vertical and horizontal pipe sections, but for purposes of this study,
emphasis will be laid on the horizontal section since this was the one used. The observation
section of the facility was placed 150 pipe diameters upstream of the last injection point to
ensure complete flow development in the horizontal section. Free air was received and
compressed by an AtlasCopco® Screw Compressor with maximum discharge pressure and
capacity of 10 barg and 400 m3•hr-1 respectively. Compressed air was discharged to a 2.5 m3
air tank before it was delivered to the test section to avoid pulsating supply. At the tank, it
was regulated to about 7 barg and fed to the test section where two gas flow meters, a 0.5-
inch (Prowirl 72F15 DN15) vortex flow meter with range of 0-20 m3/hr and a 1.5-inch (Prowirl
72F40 DN40) vortex flow meter with range from 0-130 m3/hr, both manufactured by
Endress+Hauser, were used for air metering. Air was fed to the test section through a 2-inch
steel pipe in-line with the horizontal test section.
A tank of capacity of 2 m3 was used to store heavy-oil from where it was pumped using a
variable speed progressive cavity pump with maximum capacity of 17 m 3/hr.
Endress+Hausser’s Promass 831 DN 80, coriolis flowmeter with range of 0-171 m3/hr, was
used in oil metering. The flow meter had three outputs: mass flow rate, density and viscosity.

Pet Coal (2018); 60(2): 196-209


ISSN 1337-7027 an open access journal
199
Petroleum and Coal

A refrigerated bath circulator manufactured by Thermal Fisher was used for temperature control.
Coils which were connected to the circulator were submerged in the oil tank, and hot or cold
glycol was passed through the coils to regulate the oil temperature and, hence, its viscosity.
The circulator temperature ranged from 0 to 50oC with an accuracy of ±0.01oC. Data were
acquired from the facility as raw voltage (0 – 10 V) using the data acquisition system, NI USB-
6210, and they were converted to engineering units using Labview® version 8.6.1. The 3-inch
ECT sensor was installed just after the viewing sections, which was about 155 pipe diameters
downstream of the feeding point for oil and gas. The slurry and water systems are not des-
cribed even though they are shown in the test facility schematics because they were not used
in this study.
CYL 680, a mineral oil manufactured by Total ® with density of 920 kg m-3 and viscosity of
3000 cP at 20 oC was used as the liquid phase (heavy-oil) in the dynamic test. CYL 680 and
EDM 250, which is a dielectric oil type manufactured by Rustlick ™ with viscosity 4 cP at 20 oC
and density of 801 kg m-3, were used in the bench test. For both tests, the gas used was
atmospheric air, and it was compressed in the case of the dynamic test.
The ECT device used in this study was manufactured by Industrial Tomography Systems
(ITS) in Manchester, United Kingdom. The system is made up of a sensing unit, data acqui-
sition and a computing system. The sensing unit consists of a flexible copper laminate engraved
with a predetermined electrode pattern and placed around steel pipe circumference with
internal diameters of 1 and 3 inch respectively. Computer system and data acquisition software
makes up the computing system. The software, Multi Modal Tomography Console (MMTC) has
a VC++ Graphical User Interface (GUI), which was used to view the process of interest online
and for tomography data management. Collated data could be viewed offline using another
ITS proprietary software, Toolsuite V7. The Linear Back Projection (LBP) algorithm was used
for image reconstruction in both software.
Furthermore, a gamma densitometer was used for phase fraction measurement. A fixed
single-beam gamma densitometer manufactured by Neftemer Limited made up of a gamma
source block and a Sodium Iodide (NaI) scintillation radiation detector was installed on the 3-
inch facility. A 5.5 GBq Caesium-137 radioisotope was contained in the gamma radiation
source block enclosed within a lead radiation protection shield protected by stainless steel.
The radioisotope in the gamma densitometer was a dual-energy source emitting gamma rays
in two broad photon energy levels. The 662-keV high-energy level was sourced from the
gamma radiation transmission while the lower energy level range of 100 keV–300 keV was
sourced from the scattered gamma radiation. Sodium iodide (NaI) scintillation radiation
detector was used for the measurement of two separate sets of gamma attenuation data for
the high and low energy levels with a sampling rate of 250 Hz. A detailed description of the
gamma densitometer (see pictorial view in Figure 2) used for this investigation has been pre-
sented by [13].

Figure 2. ECT static (bench) test experimental setup

Pet Coal (2018); 60(2): 196-209


ISSN 1337-7027 an open access journal
200
Petroleum and Coal

In this study, the static (bench) and dynamic (flowing) test were conducted in the 1 and 3-
inch facility using two ECT sensors of the same (test facility) size. In the static test, dielectric
oil (4 cP at 25oC) and CYL 680 oil (3,000 cP at 25oC) were used with air as the gas phase.
Liquid holdup measurements, tomographic images and raw capacitance signals were obtained
from ECT for both sensors. The sensors were first calibrated by taking the capacitance readings
for completely empty ECT sensor (full air) and recorded as the low reference since it is the
low permittivity medium; the sensor was then filled with oil and recorded again to obtain the
high reference since it was the high permittivity medium. The calibrated results were used as
references for both the static and dynamic test. In the static test, emphasis was laid on the
ability of the ECT to predict the liquid holdup measurement, visualize tomographic images and
the raw capacitance measurement. In the ECT system, tomographic images were obtained
essentially from the reconstruction of the raw capacitance data obtained by the sensors during
sensing to pixel permittivity using some form of algorithm (in this case, the Linear Back
Projection, LBP). In LBP, reconstruction was done in the Multi Modal Tomography Console
(MMTC) software. A grey level was assigned to each of the 66 sensing areas. By superimposing
the grey levels on the areas, the region where the object to be visualized was located became
enhanced. 66 element vector with each value set as 1 for object presence and 0 for none
presence was used to relate the 66 sensing areas. By multiplying the corresponding segment
vector, V, with the 66-capacitance measurement values (m1, m2……...mn), the grey level, G
was obtained. For a K number of segments, Equation (1) [14] was thus obtained.
G1 V1 m1
G2 V2 m2
. . .
. . .
. = . . (1)
.. .. ..
GK−1 VK−1 mK−1
( GK ) ( VK ) ( mK )
The equation thereby reduced to a matrix multiplication problem.
2.1. ECT referencing
Representative tomographic images results obtained in the 3-inch sensor for the air-oil
system in both the heavy-oil and air are shown in Figure 3. The images show the first tomo-
gram which is completely blue in colour indicating a completely empty sensor (100% air), that
is, 0% liquid holdup to a completely filled sensor as shown in the last tomogram, and a total
red colour indicating 100% liquid holdup (completely filled with oil). The two tomograms
mentioned were subsequently used as the references.

Figure 3. Tomographic images of different liquid holdup (%) for dielectric oil (viscosity: 3000 cP; ECT
sensor: 3 Inch)

Pet Coal (2018); 60(2): 196-209


ISSN 1337-7027 an open access journal
201
Petroleum and Coal

The tomographic images clearly indicated an increase in the red section (oil content) as the
sensor was filled with oil during the bench test. The interface between the oil and air in the
tomograms were indicated by a yellowish green colour; this region is normally considered as
aerated oil. The tomographic images showed no noticeable effect of viscosity and pipe
diameter difference on the ECT sensor. The test was conducted for both 4 cP (dielectric) and
3000 cP (CYL 680) oil with air as the gas phase using both the 1 and 3-inch sensors, and
similar results were observed for all four cases studied.
It is worth noting that even though ECT displayed the tomographic images as a
“representative” of the holdup value, the actual liquid holdup was obtained using the Maxwell’s
expression (ITS, 2000) given in Equation (2).
𝜎𝑚𝑝 𝜎2
2𝜎 + 𝜎2 − 2𝜎𝑚𝑝 −
𝜎1
𝛼= 𝜎2 (2)
𝜎𝑚𝑝 − 𝜎𝑚𝑝 + 2(𝜎1 − 𝜎2 )
𝜎1
In the equation above, 𝛼 is the volume fraction of the dispersed material, 𝜎1 , 𝜎2 and 𝜎𝑚𝑝 are
the permittivity of the continuous phase, that of the dispersed phase and the reconstructed
measured permittivity respectively.
3. Results and discussion
Results obtained from experimental investigations carried out in this study are presented
in this section. Liquid holdup plots obtained in the study are shown in Figure 4 and 5. The solid
line indicates an ideal case in which the measured and actual liquid holdup values are equal
while the dotted line indicates the error margin from the ideal case. Liquid holdup measu-
rement for 1 and 3-inch sensor at 3000 cP oil viscosity are shown in Figure 4. For both sensors,
similar results were obtained and the effect of increase in size from a 1 to 3-inch sensor was
not significant. About ±10% error margin was observed for both sensors thus showing no
significant effect of increase in pipe diameter on ECT performance. ECT performance was also
evaluated for the 4 cP oil under static condition; the results obtained were compared with the
liquid holdup measurements for 3000 cP oil and they showed no significant effect as a result
of the change in viscosity. The curved interface in the tomograms are related to the LBP
algorithm used in image reconstruction.

Figure 4. Liquid holdup comparison for 1 and 3- Figure 5. Viscosity comparison for 1-inch sensor
inch sensor (µ = 3000 cP) oil
From Figures 4 and 5, it was observed that ECT could be used to predict the liquid holdup
for actual liquid holdup values from 50% and above while overestimating liquid holdup values
below 50%. In this case, the error margin was relatively lower at the lower liquid holdup
values. This trend was observed for both 1 and 3-inch sensors as well as the two viscosities
tested. Error margins at the liquid holdup values close to the high reference (full pipe) was
seen to improve relatively to that in the mid liquid holdup values. This phenomenon is attri-
buted to the LBP algorithm used in the image reconstruction and the fact that the sensor was

Pet Coal (2018); 60(2): 196-209


ISSN 1337-7027 an open access journal
202
Petroleum and Coal

calibrated at the two terminals, the low and high reference; hence, relatively similar trends
were noticed in literature for other sensing instrument such as gamma [15].
3.1. Flow pattern identification with ECT
Annular flow pattern was modelled on the bench (static) by inserting polyvinyl chloride
(PVC) pipes of cylindrical shape with varying outer diameter (36 – 48 mm) into the oil filled
sensor in the vertical position. The low reference used here was the PVC pipe inserted into the
empty sensor while the high reference was oil. Figure 6 shows that for annular flow, the error
margin (±5%) is well within that of the stratified static test, possible reasons for error
difference were the sensor positioning and pipe material used.
Air-oil dynamic tests were conducted in the 1 and 3-inch multiphase facilities for viscosities
of 1000, 3000 and 6000 cP. ECT was used to identify flow patterns and validated by visual
observations in the sections that follow. Superficial gas and liquid velocities used in the study
were 0.3 – 9 m/s and 0.06 – 0.1 m/s, respectively.
Some representative stacked tomographic images (time series of tomographic images
taken from frames in consecutive sequence) for the 3-inch rig at 1000 cP, with Vsg values
ranging from 0.3 – 9.0 m/s and a Vso value of 0.1 m/s are shown in Table 1. The images
obtained provided information on the phase distribution in the pipeline during each flow
condition. For Vso value of 0.1 and Vsg value of 0.3 – 0.5 m/s at 1000 cP, plug flow regime
was observed. This flow pattern consisted of two characteristic tomographic images: one in
which the tomogram showed a stratified layer of red at the bottom (indicating oil) with blue
(gas) at the top section of the pipe. At certain intervals, this image almost changed completely
to red (oil), and this period indicated the passage of the entire oil with negligible amount of
gas, hence the plug body. In the stack images shown in Table 1, it is seen that oil (red colour)
completely bridged to the top as the plug body was passing. This flow pattern was classified
as plug flow based on the definition of [16], which states that plug flow is a limited case of slug
but with the liquid plug being free of entrained gas bubbles. The yellowish-green and red
colours observed at the pipe wall (in the stacked images) can be explained by the oil film
coating on the walls of the pipe housing the sensor as a result of the high-viscous property of
the oil.

Figure 6. Film and plug unit at Vso=0.1 m/s and Vsg=0.3 m/s, µl = 1000 cP

A plot of the concentration (liquid holdup) time series was also analysed for flow patterns
observed in the conditions stated above, and the plot showed intermittent behaviour,
fluctuating from a low concentration of about 0.65 to as high as 1 in some cases. Plug/slug
body were defined as such when the liquid holdup rose above 0.75 m/s. The low concentration
value indicated the film region. Here, the combine flow of air and oil ensured that the liquid
holdup value was relatively lower while the high concentration value indicated the passage of
the plug body where very little air was contained in the flow and the liquid almost completely
occupied the pipeline. The Probability Mass Function plot further confirmed the flow pattern
obtained at the aforementioned flow conditions, which clearly showed a bi-modal distribution.
ECT measurement was validated by a video recording, which was showing the plug flow
pattern observed through physical inspection and recordings. Flow patterns observed at Vso
value of 0.1 m/s and Vsg value of 0.3 -0.5 m/s at 1000 cP was found to be, actually, plug
flow. It had two characteristic units that flowed intermittently. A film unit in which the less
dense gas flowed on top of the denser liquid in a stratified pattern. This unit was disrupted at

Pet Coal (2018); 60(2): 196-209


ISSN 1337-7027 an open access journal
203
Petroleum and Coal

certain intervals by long plug bodies flowing through. The plug was observed to cover the
entire viewing section with no visible gas entrainment as shown in Table 1.
For Vso value of 0.1 and Vsg value of 0.7 – 2.0 m/s at 1000 cP, slug flow regime was
observed. The tomographic images obtained were similar to those in plug flows differing only
in the appearance of blue colour during intervals where the tomograms changed from stratified
blue (at the top) and red (at the bottom) to a wholly red colour with little quantity of blue as
presented in Figure 7.

Figure 7. Tomogram showing gas entrained in liquid slug unit at Vso=0.1 m/s and Vsg=1.0 m/s with
oil viscosity of 1000 cP

In the tomographic stack images in which the complete red image earlier seen in plug flow
was not present, a dominant red colour was observed but with visible blue albeit small in size.
This was attributed to the presence of entrained gas bubbles as the slug body passes. The
slug body was shorter and more frequent as indicated in the stacked images at these flow
conditions. The concentration time series plots and PMFs, however, did not show any signi-
ficant difference between slug and plug flows. Based on the definition of Barnea et al. [16] this
flow pattern was called slug flow. Video recordings showed a similar behaviour with plug flow
differing in the slug body being shorter, more frequent and more energetic relative to the plug
body. Gas bubbles were also observed in the slug unit as well as increased instability of the
film region. The film region instability, change in slug length and increase in flow intensity were
due to the increased kinetic energy of flow as a result of the increase in gas superficial velocity.
At Vsg values of 3 m/s in the 1000 cP test, flow regime observed in the stack images was
a transition between the slug and wavy annular flow. The oil was seen to have insufficient
energy to bridge the gas core thus forming a kind of wavy pattern around the gas as shown
in Figure 8. This classification was chiefly subjective, but the PMF plot laid further credence to
this as the hitherto, bimodal distribution in plug and slug flows merges partially with each
other. This regime was defined as blow-through-slug or pseudo-slug based on the definition
of Wong and Yau [17]. In the video recordings, a further increase in Vsg caused the gas kinetic
energy to acquire even more momentum resulting in slugs requiring relatively larger energy
to bridge the liquid film. It is similar to the wavy annular flow differing only in the fact that
the gas velocity is not high enough to sweep the liquid to the top of the pipe.

Figure 8. Tomogram showing Pseudo-Slug body at Vso=0.1 m/s and Vsg=3.0 m/s for oil viscosity of 1000 cP

Finally, at Vsg value of 7 – 9 m/s with oil viscosity of 1000 cP and 5 – 9 m/s, the stack
tomographic images showed a wavy flow structure between the gas-oil interface, and the
tomograms were similar to the pseudo-slug flow patterns but with relatively low penetration
height into the gas zone. The concentration time series also showed a high fluctuation in
behaviour with the PMF showing unimodal distribution. In the video recordings, the increased
Vsg value caused increase in dissipated energy along the flow resulting in large amplitude of
waves at the oil-gas interface. The top of the pipe wall was significantly wetted by oil with gas
continually sweeping liquid at the interface to the top of the pipe. Most of the oil in flow,
however, remained at the bottom of the pipe (due to gravity) with gas predominantly flowing
at the centre and thin oil film at the top. The PMF, liquid holdup plot and the tomographic

Pet Coal (2018); 60(2): 196-209


ISSN 1337-7027 an open access journal
204
Petroleum and Coal

images obtained from ECT showed promising results in the capability of ECT to visualize high-
viscous oil and gas flows as validated by the video recordings obtained during the study.
Flow pattern maps from the study are shown in Figure 9 and 10. From the figures, an
increase in viscosity was seen to result in a corresponding increase in the slug/plug flow
pattern region. This was found to agree with the conclusion drawn by Márquez and Trujillo [4]
and Gokcal et al. [4]. Also, an increase in the oil velocity had similar effects on the flow pattern
within the range of superficial oil and gas velocities tested.

Figure 9. Flow pattern map for 1-inch multiphase Figure 10. Flow pattern map for 3-inch multiphase
facility facility

3.2. Validation of flow patterns obtained by ECT with gamma densitometer


A representative time series and PDF plots of the time varying liquid holdup measurement
of ECT and Gamma densitometer measurements are presented in Table 2. Both data obtained
from the 3-in pipe ID horizontal test facility are shown for 20 s each to facilitate a clear
representation of the results.
At gas superficial velocity of 0.3 m/s, the ECT time series plots showed fluctuations in
waveforms. This was an indication of the plug flow pattern in this flow condition. The crests of
the waveforms indicated elongated liquid body passage, and each crest was observed to be a
result of the sudden increase in the liquid holdup as the liquid bridged the entire pipe cross
section. The troughs, which were indications of the film region passage, were characterized
by relatively low liquid holdup values, which were explained by the increased void fraction as
the liquid bridge hitherto prevalent in the elongated liquid body exited the observation section.
Similar waveform characteristics were observed in the Gamma waveform signals shown in
Table 2. PDF plots for this flow condition showed a bimodal distribution having two
characteristic peaks with one at the higher holdup value indicating the slug liquid body while
the other was representing the slug film.
At Vsg of 3.0 m/s, the flow pattern entered its transition region (pseudo-slug), and it was
observed that both the ECT and gamma time series plots indicated less frequent crests, which
was a result of the reduced intermittency in the flow. PDF plots from both ECT and Gamma
indicated different modal distributions. This is as a result of the increased transient nature of
the transition flow pattern and/or limitations of the single beam Gamma device used in this
case, which was impeding its ability to capture clearly the features of the transition region.
For Vsg values of 9 m/s, both Gamma and ECT time series analysis shows a chaotic
behaviour in the time series waveforms with no distinguishable crest or trough. The PDF, at
this flow condition, showed a unimodal distribution which clearly suggested the non-
intermittency of the flow pattern.
It can therefore be inferred that within the experimental test matrix and test conditions of
this study, flow patterns identified by ECT were similar to those identified by Gamma
densitometer.

Pet Coal (2018); 60(2): 196-209


ISSN 1337-7027 an open access journal
205
Table 1. EC T results for oil viscosity 1000 cP oil superficial velocity 0.1 m/s with different gas superficial velocities ( 3-inch EC T sensor)

Vsg ECT Time Series Plots ECT PDF Plots Stacked Images

0.3 m/s

Plug Flow

206
3.0 m/s

Pseudo-Slug Pseudo-Slug Flow

9.0 m/s

Wavy Annular Flow


Table 2. EC T results and Gamma Densitometer validation for oil viscosity 1000 cP oil superficial velocity 0.1 m/s with different gas superficial
velocities (3-inch EC T sensor)
Vsg ECT Time Series Plots Gamma Time Series Plots ECT PDF Plots Gamma PDF Plots

0.3
m/s

3.0

207
m/s

9.0
m/s
Petroleum and Coal

Figure 11. Plot of liquid holdup against Vsg at Figure 12. Plot of liquid holdup predicted by “No-Slip”
different Vso for 1- and 3-inch ECT sensor equation and ECT sensor against Vsg (m/s) for the 3-
inch test facility at oil viscosity of 3000 cP
Figure shows time averaged liquid holdup measurement obtained for 60 s from ECT in both
the 1- and 3-inch test facility for viscosities ranging from 1000 – 7500 cP. The data plot
showed a general trend of reduction in liquid holdup value as the gas superficial velocity was
increasing. This was expected to be so in order to satisfy the continuity equation. As the gas
superficial velocity was increased, the gas was observed to occupy more cross sectional area
in the pipe resulting in a corresponding reduction of the liquid holdup in the cross sectional
area of the pipe. An increase in the viscosity of the oil resulted in an increase in the liquid
holdup at fixed Vsg and Vso. This was due to the fact that an increase in viscosity resulted in
an increase in the shear stress (resistance to flow) of the oil. A reduction in the amount of
entrained gas in the slug body (for slug flows) was also observed visually as well as from the
time series trace in ECT. The gas kinetic energy hitherto sufficient to displace a particular
amount of oil then became insufficient, thereby resulting in more liquid remaining in flow and
thus increased liquid holdup. A comparison of the liquid holdup obtained from ECT to that of
the “no slip” liquid holdup as shown in Figure was found to show a similar trend. However,
the liquid holdup values were significantly different; this was due to the “no slip” assumptions
in addition to the increased liquid viscosity tested.
4. Conclusion
ECT provided good tomographic images and liquid holdup measurements under static
(bench) conditions. The results obtained showed that the measurement of ECT was accurate
to within ±10% of the actual liquid holdup under stratified static conditions. Under dynamic
(flowing) conditions, ECT was able to provide good tomographic images for the distribution of
the phases in flows - plug, slug, pseudo-slug (blow-through-slug) and annular flows – that
were observed. PMF plots of the liquid holdup data and the time series were used to further
identify the different regimes. Flow patterns obtained were validated with the aid of video
recordings and gamma ray densitometer. Liquid viscosity and pipe diameter had no significant
effect under static conditions on ECT performance. However, for dynamic tests, increase in
liquid viscosity led to a corresponding increase in the oil film coatings on the sensor walls.
Shorter slugs and increased slug frequency with increase in liquid viscosity was clearly
visualized by ECT. This was found to be in agreement with the studies performed by Márquez
and Trujillo [5] and Gokcal et al. [4] for high viscous oil-gas flows.
Acknowledgement
The authors acknowledge the contribution of the Laboratory Manager, Technicians,
Research Staff Members and Students in the Oil and Gas Engineering Centre of Cranfield
University for their help and support rendered in the course of this research work.

Pet Coal (2018); 60(2): 196-209


ISSN 1337-7027 an open access journal
208
Petroleum and Coal

References
[1] Yang W and Peng L. Image Reconstruction Algorithms for Electrical Capacitance Tomography.
Meas. Sci. Technol., 1996; 14(1):
[2] Dyakowski T. Process tomography applied to multi-phase flow measurement. Meas. Sci.
Technol., 1996; 7. 343–353.
[3] Beck MS, Byars M, Dyakowski T, Waterfall R, He R, Wang SJ, Yang WQ. Principles & Industrial
Applications of Electrical Capacitance Tomography. Meas. Control, 1997; 30(7): 197–200.
[4] Gokcal B, Al-Sarkhi A, Sarica C, and Alsafran EM. Prediction of Slug Frequency for High Viscosity
Oils in Horizontal Pipes,” in SPE Annual Technical Conference and Exhibition, 2009.
[5] Márquez J and Trujillo J. Overview : Slug-Flow Characterization for Heavy-Oil Fields. in
Proceeding of SPE Latin American and Caribbean Petroleum Engineering Conference, 2010, no.
December, pp. 1–3.
[6] Gamio J, Castro J, Rivera L, Alamilla J, Garcia-Nocetti F, and Aguilar L. Visualization of Gas-Oil
Two-Phase Flows in Pressurized Pipes Using Electrical Capacitance Tomography. Flow Meas.
Instrum.,2005; 16: 129–131.
[7] Zhang L and Wang H. Identification of oil–gas two-phase flow pattern based on SVM and
electrical capacitance tomography technique. Flow Meas. Instrum., 2010; 21(1): 20–24.
[8] Baker G, Azzorpadi B, and Clark W. Transients in stratified/slug flow – observations using
electrical capacitance tomography. in International Conference on Multiphase Technology,
2003.
[9] Taitel Y and Dukler AE. A model for predicting flow regime transitions in horizontal and near
horizontal gas-liquid flow. AIChE J.,1976; 22(1): 47–55.
[10] Isaksent O, Dicot A, and Hammer EA. A Capacitance-Based Tomography System for Interface
Measurement in Separation Vessels. Meas. Sci. Tech, 1994; 5: 1262–1271.
[11] Warsito W and Fan L. Flow Structures in Gas–Liquid & Gas–Liquid–Solid Flow Systems Using
Electrical Capacitance Tomography. Chem. Eng. Sci., 2001; 56: 5871-5891.
[12] Jeanmeure L, Dyakowski T, William B, and Zimmerman W. Direct Fow-Pattern Identification
Using Electrical Capacitance Tomography. Exp. Therm. Fluid Sci., 2002; 26: 763–777.
[13] Archibong A, Zhao Y, and Yeung H. Comparison of electrical capacitance tomography &
gamma densitometer measurement in viscous oil-gas flows. AIP Conf. Proc., 2014; 1592(1).
[14] Plaskowski A, Beck MS, Thorn R, and Dyakowski T. Imaging industrial flows: Application of
electrical process tomography. Bristol and Philadelphia: Institute of Physics Publishing, 1995.
[15] Åbro E and Johansen G. Improved void fraction determination by means of multibeam gamma-
ray attenuation measurements Flow Meas. Instrum., 1999; 10(2): 99–108.
[16] Barnea D, Shoham O, Taitel Y and Dukler AE. Flow pattern transition for gas-liquid flow in
horizontal and inclined pipes. Comparison of experimental data with theory. Int. J. Multiph.
Flow, 1980; 6(1976): 217–225.
[17] Wong T and Yau Y. Flow Patterns in Two-Phase Air-Water Flow. Int. Commun. Heat Mass Transf.,
1997; 24(1): 111–118.

To whom correspondence should be addressed: Yahaya D. Baba, Department of Chemical and Petroleum Engineering,
Afe Babalola University,Ado-Ekiti, Nigeria, y.baba9550@gmail.com

Pet Coal (2018); 60(2): 196-209


ISSN 1337-7027 an open access journal
209